© ESO, 2012

1. Introduction

A critically important region of the astrophysical spectrum is the hard X-ray band, from 15 keV to 200 keV, which has been explored in detail by the two satellites, INTEGRAL (Winkler et al. 2003) and Swift (Gehrels et al. 2004), which carry the instruments IBIS (Ubertini et al. 2003) and BAT (Barthelmy 2004) operating in the 20−200 keV band. These spacecrafts permit a study of the processes taking place in this observational window providing a deep look into the physics of hard X-ray sources.

These telescopes operate in a complementary way, as the first concentrates on mapping the Galactic plane, while the second mainly covers the high Galactic latitude sky, so that together they provide the best sample of objects yet selected in the hard X-ray domain. So far, both instruments have detected a large number of both known and new objects, discovered new classes of sources, and allowed us to find and study highly absorbed objects. In particular, the nature of many objects detected above 20 keV by both satellites is often unknown, the sources being optically unclassified and their types only being able to be inferred based on few available X-ray or radio observations.

Optical follow-up of these sources is therefore mandatory. In particular, the optical spectra can provide not only an accurate source classification, but also fundamental parameters that together with multiwaveband studies, for example in the soft X-ray band, can provide information on these newly detected objects.

In this paper, we focus on the X-ray and optical follow-up work of a number of objects with unknown classifications and/or redshifts, reported in the 39 month Swift/BAT survey catalogue (Cusumano et al. 2010a). We note that the identifications of the present paper are also reported in the Palermo 54 month catalogue (Cusumano et al. 2010b), and that preliminary classifications were given by us via private communication. Our aim is indeed to perform a systematic study of unidentified Swift /BAT objects starting with the 39 month surveys and continuing with the identifications of those of the 54 months catalogue (Parisi et al., in prep.).

This survey covers 90% of the sky down to a flux limit of 2.5 × 10^-11 erg cm^-2 s^-1 and 50% of the sky down to a flux limit of 1.8 × 10^-11 erg cm^-2 s^-1 in the 14−150 keV band. It lists 754 sources, of which 69% are extragalactic, 27% are galactic, and 4% are of unknown type.

From this BAT survey, we selected a sample of 29 objects either without optical identification, or that had not been well studied, or without published optical spectra. For all these sources, we first performed the X-ray data analysis to reduce the source positional uncertainty from arcmin-to arcsec-sized radii and derive information on the main spectral parameters (photon index, column density, and 2−10 keV flux). Within the reduced X-ray error boxes, we then identified the putative optical counterpart to the BAT object and performed optical spectroscopic follow-up work. Following the method applied by Masetti et al. (2004, 2006a,b, 2008, 2009, 2010, 2012) and Parisi et al. (2009), we determined the nature of all selected objects and discussed their properties. A preliminary classification of these sources was given in the Palermo 54 month BAT catalogue (Cusumano et al. 2010b), while here we publish for the first time the optical spectra and detailed optical information (see Tables 5–7).

We also checked for the presence of peculiar sources, such as Compton thick AGN, absorbed Seyfert 1’s and unabsorbed Seyfert 2’s, using the diagnostic method of Malizia et al. (2007), and finally used the plot of Maiolino et al. (2001) to verify a possible mismatch between the X-ray gas absorption and the optical dust reddening.

The paper is structured as follow: in Sect. 2, we report information on the soft X-ray data analysis; in Sect. 3, we describe the optical observations, the telescope employed, and provide information on the data reduction method used. Section 4 reports and discusses the main optical results (line fluxes, distances, Galactic and local extinction, central black hole masses etc.). In Sect. 5, the X-ray and the optical results are compared in view of the object classification and gas versus dust absorption. In Sect. 6, we summarize the main conclusions of our work.

2. X-ray data analysis

We now provide general information about the X-ray data analysis performed for the 29 objects of our sample, to obtain indications of the X-ray counterpart of the BAT object, provide its position with arcsec accuracy (see Table 1), and finally study its spectral properties in the 2−10 keV band (see Table 3).

For 24 of the 29 objects, we used X-ray data acquired with the X-ray Telescope (XRT, 0.3−10 keV, Burrows et al. 2004) onboard the Swift satellite. The XRT data reduction was performed using the XRTDAS standard data pipeline package (xrtpipeline v. 0.12.6), to produce screened event files. All data were extracted in only the photon counting (PC) mode (Hill et al. 2004), adopting the standard grade filtering (0–12 for PC) according to the XRT nomenclature. Depending on the source nature (bright or dim), we either used the longest exposure or added multiple observations to enhance the signal-to-noise ratio (S/N). For each BAT detection, we then analysed, with ximage v. 4.5.1, the 3−10 keV image of interest (single or added over more XRT pointings) to search for sources detected (at a confidence level >3σ) within the 90% Swift /BAT error circles; this 3−10 keV image choice ensured that we selected the hardest sources, hence the most likely counterparts to the BAT objects. We estimated the X-ray positions and relative uncertainties using the task xrtcentroid v.0.2.9.

For 5 sources, we instead used X-ray data acquired with the pn X-ray CCD camera on the EPIC instrument onboard the XMM-Newton spacecraft (Strüder et al. 2001) in order to determine more reliably the spectral properties of those objects with low quality XRT data. These data were processed using the Standard Analysis Software (SAS) version 9.0.0 employing the latest available calibration files. Only patterns corresponding to single and double events (PATTERN ≤ 4) were taken into account and the standard selection filter FLAG = 0 was applied. For each source, we analysed a single observation, the longest in terms of exposure, to achieve higher quality statistics (see Table 2 for the observation (IDs)). In each case, we searched the XMM-Newton EPIC-pn images for X-ray sources that fell inside the Swift/BAT error circles; to avoid false associations with dim and soft objects, we only inspected higher energy (>4 keV) images in this case.

In 28 out of 29 objects analysed, a single object was detected within the BAT 90% positional uncertainty; the exception was PBC J0041.6+2534 for which we found one soft X-ray source located just outside the 90% error circle and none within it. Nevertheless, we consider it as the likely counterpart to the BAT source.

Next, we analysed the X-ray spectrum of each object. Events for spectral analysis were extracted within a circular region of radius 20′′, centred on the source position for XRT¹ and choosing instead the radius corresponding to the highest Signal-to-Noise ratio for EPIC-pn. The background was taken from empty regions close to the X-ray source of interest, using circular regions with different radii for XRT data, to ensure an evenly sampled background. In the case of EPIC-pn, data we adopted instead an 80′′ radius for all the 5 analysed sources. The XRT spectra were then extracted from the corresponding event files using the xselect v.2.4 software and binned using grppha in an appropriate way, so that the χ² statistic could be applied. We used version v.011 of the response matrices and created the relative ancillary response file arf using the task xrtmkarf v.0.5.6.

For XMM spectra, ancillary response matrices (ARFs) and detector response matrices (RMFs) were generated using the XMM-Newton SAS tasks arfgen and rmfgen, respectively. In general, the spectral channels were rebinned to achieve a minimum of 20 counts in each bin (see however Table 3).

The energy band used for the spectral analysis, which was performed with XSPEC v.12.6.0, depends on the statistical quality of the data and typically ranges from 0.3 keV to ~6 keV for XRT and from 0.5 keV to 12 keV for EPIC-pn.

In the first instance, we adopted, as our basic model, a simple power-law passing through the Galactic (Dickey & Lockman 1990) and (when required) intrinsic absorption.

If this baseline model was insufficient to fit the data, we introduced additional spectral components (i.e. a black body, second-power law, or Gaussian line) as required according to the F-test statistics.

We note that in some cases owing to the limited statistical quality of the XRT data we fixed the photon index to 1.8 (which is the canonical value for AGN; e.g., Mushotzky et al. 1993) to estimate the intrinsic column density and/or the X-ray flux.

We are aware that in a few cases the statistical quality of the data is such that the obtained parameter values must be interpreted with caution. Nevertheless, they provide some indications of the source properties, i.e. whether a source is likely to be absorbed or not.

The results of the X-ray spectral analysis are reported in Table 3, where we list for each object the Galactic column density, the power-law photon index, the column density in excess to the Galactic value, the reduced χ² of the best-fit model, the 2–10 keV flux, and the 20–100 keV flux² additional spectral parameters, if they were required, are reported in the notes at the end of the table. Here and in the following all quoted errors correspond to a 90% confidence level for a single X-ray parameter of interest (Δχ² = 2.71).

We note that for a limited number of our sources, X-ray data were previously published. Winter et al. (2009) discuss the spectra of PBC J0641.3+3251 and PBC J0356.9-4040 while Noguchi et al. (2009) analysed the XMM-Newton of PBC J0919.9+3712. Finally, PBC J2148.2-3455 has been extensively studied at X-ray energies using various instruments, such as XMM-Newton and Chandra (see for example González-Martín et al. 2009; Levenson et al. 2005), but no XRT data have been yet presented. Overall, we find good agreement with these previous studies.

3. Optical spectroscopy

We describe the optical follow-up studies that we performed for all 29 objects. We list in Table 1 the coordinates of the optical counterparts obtained from the 2MASS catalogue³ (Skrutskie et al. 2006). In all but one case, these optical counterparts coincide in position with a galaxy, which immediately suggests that the majority of the objects listed in Table 1 are active galaxies (their name as derived by NED⁴ is reported in Tables 6 and 7). The only exception is PBC J0826.3-7033, which is listed in NED and SIMBAD⁵ as an unidentified X-ray source; our optical follow-up study located this source at z = 0, thus demonstrates that it is not an extragalactic object but rather a Galactic high-energy source.

Of the 28 AGN, 10 objects already have a known redshift, although these are not supported by published optical spectra, 9 have redshifts obtained from the Sloan Digital Sky Survey⁶ (SDSS, Adelman-McCarthy et al. 2007) and the Six-degree Field Galaxy Survey (6dFGS; Jones et al. 2004) archives and 9 have redshifts that were derived from our spectroscopic observations and therefore published here for the first time. In one case (a 6dF spectrum), our result differs significantly from the one reported in the literature (see Sect. 4), suggesting that it is important to confirm published redshift values especially for newly discovered objects.

Concerning the optical classes for all our sample, 12 objects had an optical class that had already been reported in the Veron-Chetty & Veron 13th catalogue edition (V&V13, Veron-Cetty & Veron 2010 and references therein) and/or in NASA/IPAC extragalactic database (NED). Nevertheless, we chose to report data on these 12 sources for a number of reasons: in a few cases, our classification is different or more detailed than the one reported in the literature, and for a couple of sources different authors provide contradictory results, thus the classification is ambiguous. For the remaining 17 objects, we reported the optical class supported by optical spectra for the first time.

In Table 1, the detailed log of all optical measurements is also reported: we list in Col. 7 the telescope and instrument used for the observation, while the characteristics of each spectrograph are given in Cols. 8 and 9. Column 10 provides the observation date and the UT time at mid-exposure, while Col. 11 reports the exposure times and the number of spectral pointings.

For 20 sources, the following telescopes were used for the optical spectroscopic study presented here:

• the 1.52 m “Cassini” telescope of the Astronomical Observatory of Bologna, in Loiano, Italy;

• the 1.9 m “Radcliffe” telescope of the South African Astronomical Observatory (SAAO) in Sutherland, South Africa;

• the 2.1 m telescope of the Observatorio Astrónomico Nacional in San Pedro Martir, Mexico;

• the 3.58 m New Technology Telescope (NTT) at the ESO-La Silla Observatory, Chile.

The data reduction was performed with the standard procedure (optimal extraction, Horne 1986) using IRAF⁷. Calibration frames (flat fields and bias) were taken on the day preceding or following the observing night. The wavelength calibration was obtained using lamp spectra acquired soon after each on-target spectroscopic acquisition. The uncertainty in the calibration was ~0.5 Å in all cases; this was checked using the positions of background night-sky lines. Flux calibration was performed using catalogued spectrophotometric standards. Objects with more than one observation had their spectra stacked together to increase the S/N.

Additional spectra (that is, 9 out of 29) were retrieved from two different astronomical archives: the SDSS and the 6dFGS⁸. As the 6dFGS archive provides spectra that have not been flux calibrated, we used the optical photometric information in Jones et al. (2005) and Doyle et al. (2005) to calibrate the 6dFGS data presented here.

The identification and classification approach we adopt in the analysis of the optical spectra is the following: for the emission-line AGN classification, we used the criteria of Veilleux & Osterbrock (1987) and the line-ratio diagnostics of Ho et al. (1993, 1997) and Kauffmann et al. (2003) to distinguish among the Seyfert 2, starburst galaxies, Hii regions, and low-ionization nuclear emission-line regions (LINERs; Heckman 1980). In this last class, some lines ([Oii]_λ3723, [Oi]_λ6300, and [Nii]_λ6584) are stronger than in typical Seyfert 2 galaxies; the permitted emission-line luminosities are weak; and the emission-line widths are comparable with those of type 2 AGNs. In particular, as mentioned in Ho et al. (1993), all sources with [Oii] > [Oiii], [Nii]/H_α > 0.6, [Oi] > 1/3 [Oiii] can be considered LINERs (see Table 7). For the subclass assignation to Seyfert 1 galaxies, we used the H_β/[O iii]λ5007 line flux ratio criterion presented in Winkler et al. (1992). Moreover, the criteria of Osterbrock & Pogge (1985) allowed us to discriminate between “normal” Seyfert 1 and narrow-line Seyfert 1 (NLS1): the latter are galaxies with a full width at half-maximum (FWHM) of the H_β line lower than 2000 km s^-1, with permitted lines which are only slightly broader than their forbidden lines, with a [Oiii]_λ5007/H_β ratio <3, and finally with evident Feii and other high-ionization emission-line complexes.

We note that the spectra of all extragalactic objects are not corrected for starlight contamination (see, e.g., Ho et al. 1993, 1997), because of their limited S/N and spectral resolution. However, this does not affect our results and conclusions.

To estimate the E(B − V) local optical absorption in our AGN sample, when possible, we first dereddened the H_α and H_β line fluxes by applying a correction for the Galactic absorption along the line of sight to the source. This was done using the galactic colour excess E(B − V)_Gal given by Schlegel et al. (1998) and the Galactic extinction law obtained by Cardelli et al. (1989). We then estimated the colour excess E(B − V)_AGN local to the AGN host galaxy by comparing the intrinsic line ratio and corrected that for Galactic reddening using the relation for type 2 AGNs derived from Osterbrock (1989)

$\mathit{E}\mathrm{(}\mathit{B}\mathrm{-}\mathit{V}\mathrm{)}\mathrm{=}\mathit{a}\mathrm{Log}\left(\frac{{\mathrm{H}}_{\mathit{\alpha }}\mathit{/}{\mathrm{H}}_{\mathit{\beta }}}{\mathrm{(}{\mathrm{H}}_{\mathit{\alpha }}\mathit{/}{\mathrm{H}}_{\mathit{\beta }}{\mathrm{)}}_{\mathrm{0}}}\right)\mathrm{·}$

For type 1 objects, where the H_α is strongly blended with the forbidden narrow [Nii] lines, it is not easy to obtain a reliable H_α/H_β estimate. In these cases, we used the H_γ/H_β ratio, taking into account that H_γ may also be blended with the [O iii]_λ4363 line. In the above relation, H_α/H_β is the observed Balmer decrement, (H_α/H_β)₀ is the intrinsic one (2.86), and a is a constant with a value of 2.21. When instead we used the H_γ, we adopted the same relation described before, using an intrinsic (H_γ/H_β)₀ of 0.474 with an a value of −5.17.

Finally, we estimated the mass of the central black hole for a few type 1 AGN found in the sample⁹. The method used here follows the prescription of Wu et al. (2004) and Kaspi et al. (2000), where we used the H_β emission line flux, corrected for the Galactic colour excess (Schlegel et al. 1998), and the broad-line region (BLR) gas velocity (v_FWHM). Using Eq. (2) of Wu et al. (2004), we estimated the BLR size, which is used with v_FWHM in Eq. (5) of Kaspi et al. (2000) to calculate the AGN black hole mass. The results are reported in Table 8.

To derive the distance of the only compact Galactic X-ray source of our sample, we used the distance modulus assuming an absolute magnitude M_V ~ + 9 and an intrinsic colour index (V − R)₀ ~ 0 mag (Warner 1995). Although this method basically provides an order-of-magnitude value for the distance of this Galactic source, our past experience (Masetti et al. 2004, 2006a,b, 2008, 2009, 2010) tells us that this estimate is in general correct to within 50% of the refined value subsequently determined with more precise approaches. To calculate instead the luminosity distances of the 28 galaxies in the sample, we considered a cosmology with h₀ = 70 km s^-1 Mpc^-1, Ω_Λ = 0.7, and Ω_m = 0.3 and used the Cosmology Calculator of Wright (2006).

Fig. 1 Spectrum (uncorrected for the intervening Galactic absorption) of the optical counterpart of the Galactic CV PBC J0826.3-7033.

Fig. 2 Spectra (not corrected for the intervening galactic absorption) of the optical counterpart of PBC J0041.6+2534, PBC J0100.6−4752, PBC J0122.3+5004, PBC J0140.4−5320, PBC J0248.9+2627, PBC J0353.5+3713, PBC J0356.9−4040, PBC J0503.0+2300, PBC J0543.6−2738, PBC J0544.3+5905, PBC J0623.8−3212 and PBC J0641.3+3251, PBC J0759.9+2324, PBC J0814.4+0421 and PBC J0919.9+3712.

Fig. 3 Spectra (not corrected for the intervening galactic absorption) of the optical counterpart of PBC J0954.8+3724, PBC J1246.5+5432, PBC J1335.8+0301, PBC J1344.2+1934, PBC J1345.4+4141, PBC J1439.0+1413, PBC J1453.0+2553, PBC J1506.6+0349, PBC J1546.5+6931, PBC J1620.3+8101, PBC J2148.2−3455, PBC J2333.9−2343 and PBC J2341.9+3036.

4. Optical classification

We discuss the optical classifications found and highlight the most interesting or peculiar objects discovered. The B magnitudes if not otherwise stated, are extracted from the Linked Extragalactic Database and Archives (LEDA, Prugniel et al. 2005) and the R magnitudes from the USNO-A2.0 catalogue (The United States Naval Observatory, Monet et al. 2003).

4.1. Galactic object

PBC J0826.3−7033 is the only source that displays emission lines of the Balmer complex (up to at least H_ϵ), as well as He i and He ii, consistent with z = 0, indicating that this object lies within our Galaxy (see Fig. 1). The analysis of all these optical features indicates that this source is a cataclysmic variable (CV, see Table 1). The He ii_λ4686/H_β equivalent width (EW) ratio, which is smaller than 0.5, and the EW of these two emission lines (only the H_β EW is larger than 10 Å) point out that this source is likely a non-magnetic CV (see Warner 1995, and references therein). The H_α to H_β flux ratio is ~1.5 allowing us to assume that the absorption along the line of sight is negligible. This is in line with the hydrogen column density value obtained from the X-ray spectral analysis (see below). The source was previously detected at soft X-ray energies, being listed for example in the Rosat Bright source catalogue (Voges et al. 1999); as for many other CVs, PBC J0826.3−7033 is also located at relatively high Galactic latitudes, i.e. 18 degrees above the Galactic plane.

We also estimated its distance to be 90 pc, i.e. relatively close, assuming no Galactic extinction along the line of sight. The X-ray spectrum is best fitted with a bremsstrahlung model (see Table 4 for more information). At the estimated distance, the 2−10 keV source luminosity is around 2 × 10³⁰ erg s^-1, which is relatively low compared to the CVs so far detected in hard X-rays (Landi et al. 2009). Finally, we estimated the mass of the white dwarf using Eqs. (5) and (6) of Patterson & Raymond (1985). Using the bremsstrahlung temperature of our model (see Table 4) and a 0.2−4 keV luminosity of ~2.5 × 10³⁰ erg s^-1, we obtained a value of about 0.4 M_⊙.

4.2. Extragalactic objects

The results of our optical study of extragalactic sources are reported in Tables 6 and 7, where for each source we list the H_α, H_β, and [Oiii] fluxes, the classification, the redshift estimated from the narrow lines, the luminosity distance given in Mpc, the Galactic colour excess and the colour excess local to the AGN host. All the extragalactic optical spectra are displayed in Figs. 2 and 3. Of the 28 active galaxies found, 7 have strong redshifted broad and narrow emission-lines that are typical of Seyfert 1 galaxies, while the remaining 21 display only the strong and redshifted narrow emission-lines typical of Seyfert 2 galaxies (for the subclass classification see Tables 6 and 7). As reported before, some sources have a preliminary classification and/or redshift in the Palermo 54 month BAT catalogue (Cusumano et al. 2010b), while here we publish for the first time their optical spectra and the corresponding information.

4.2.1. Redshifts

We confirm the redshift estimates reported in NED, V&V13, and Ciroi et al. (2009) for 19 AGN. In one case (PBC J0623.8-3212), we obtain a different redshift (0.035) from the one (0.022) already available, despite both redshifts being extracted from the same 6dF spectrum. The origin of this discrepancy is unclear. For the remaining 9 sources, we report the redshifts derived from low-resolution optical spectra for the first time (see Figs. 2 and 3). Redshifts values are in the range 0.008−0.075, which means that our sources are all located in the local Universe.

All redshifts were estimated using the [Oiii] narrow emission line and when this line was unavailable, from either the forbidden narrow emission lines or absorption features.

4.2.2. Optical class

For the first time, we also provide the classification of 15 sources in the sample. For the remaining 14 objects, our results only partially (50%) agree with the classifications listed in the literature.

For 5 objects (PBC J0100.6-4752, PBC J0356.9-4040, PBC 0503.0+2300, PBC 0814.4+0421, and PBC J1335.8+0301) we find a different AGN type from the one already reported. The change differs slightly for the first two sources that are now classified as Seyfert 2 galaxies rather than Seyfert 1.8−1.9 (Baumgartner et al. 2008; Winter et al. 2009); the issue is instead more important for the remaining 3 objects. PBC 0503.0+2300 and PBC 0814.4+0421 move from a Seyfert 1 classification (Cusumano et al. 2010a) to a Seyfert 1.5 and to a narrow line Seyfert 1 (NLS1), respectively. PBC J1335.8+0301 shifts from a type 1 (V&V13) to a type 2 class.

PBC J2148.2-3455 (also named NGC 7130 or IC 5135) is a known AGN, but with multiple classifications. Phillips et al. (1983), Heisler et al. (1997), and Vaceli et al. (1997) classified it as a Seyfert 2 galaxy, Thuan (1984) and Veilleux et al. (1997) assigned it a LINER type, while NED and V&V13 list it as a Seyfert 1.9. We confirm the Seyfert 2 nature of PBC J2148.2-3455 and suggest that the differences in the optical classification may simply reflect the different contributions from the starburst emission in the observations.

Baldwin et al. (1981) classified PBC J2333.9-2343 (also PKS 2331-240) as a Seyfert 2 galaxy, Radivich & Kraus (1971) instead reported it as a narrow-line radio galaxy, and Andrew et al. (1971) classed it as a quasi-stellar object, but with no redshift. Bolton (1975) defined the object morphology as an elliptical galaxy; as a result, NED classified it as a Seyfert of unclear type. We were finally able to classify the object as a Seyfert 2, thus confirming the original classification.

4.2.3. Peculiar sources and discussion

Within the sample of type 1 AGNs listed in Table 6, we find that one is a Seyfert 1, five are AGNs of intermediate type (1.2−1.9), and one is a NLS1.

In terms of the unified model, intermediate Seyferts have been interpreted as objects in which our line of sight progressively intercepts the obscuring torus starting from its outer edge. However, this is not the only possible interpretation as intermediate classifications may be related to other phenomena, such as an intrinsically variable ionizing continuum. For example, a source that would normally appear as a Seyfert 1 can be classified as an intermediate-type AGN when found in a low flux state (Trippe et al. 2010). Insights into the properties of our intermediate Seyferts can help in discriminating between the various scenarios (see next section).

PBC J0814.4+0421 deserves a special mention among type 1 AGN. LEDA 023094, which is its optical counterpart with magnitude B = 15.5 and redshift of 0.027, displays optical features typical of NLS1 (see Sect. 3). These sources are rare among hard X-ray selected objects since their fraction is only 5% of all AGN and 10% of type 1 Seyferts (Panessa et al. 2011). A possible interpretation of the peculiar observational properties of NLS1 is that these systems are accreting close to their Eddington limit, implying that, compared to typical Seyfert 1 galaxies, they should host black holes with lower masses (M_BH ≤ 10⁷ M_⊙). PBC J0814.4+0421 indeed has the lowest black hole mass among the four objects for which this parameter has been estimated in this work; the value obtained is also compatible with those of other hard X-ray selected NLS1 (Panessa et al. 2011). Interestingly, PBC J1453.0+2553, the only object classified as a pure Seyfert 1 has by far the highest black hole mass observed among our small sample of type 1 AGN.

A large fraction of our extragalactic objects belong to the type 2 AGN class; this is not unexpected, as hard X-ray surveys are very efficient in discovering this type of galaxies. Among type 2 AGNs, there are also a few interesting cases, such as LINERs.

PBC J0041.6+2534 and PBC J1344.2+1934, for example, are located in an intermediate region between Seyfert 2’s and LINERs in the diagnostic diagrams (Ho et al. 1993; Kauffmann et al. 2003); because of this proximity to Seyfert 2 galaxies, both are treated here as type 2 AGNs. PBC J0353.5+3713 is instead a pure LINER, but since it displays only narrow emission lines in the optical spectrum it is also considered as a type 2 AGN. Interestingly, all three objects with LINER signatures are absorbed in X-rays, thus confirming their similarity with type 2 AGN.

As a final remark, we note that all our AGN have X-ray spectra typical of their class; that is, a simple power law (either intrinsically absorbed or not) plus in many cases an extra soft component that can be parameterized by either a second power law (having the same photon index as the primary component) or a black body model. In only 3 objects do we detect emission lines compatible with neutral (in two cases) and ionized (only in one case) iron¹⁰. We do not comment on these spectra further, but we use the information on the intrinsic column density to compare in the following sections the optical versus X-ray classification and discuss the optical (dust) versus X-ray (gas) absorption.

5. X-ray versus optical classification and absorption

The unification scheme states that every AGN is intrinsically the same object, namely an accreting supermassive black hole surrounded by an obscuring torus. Depending on how the observer views the central engine, an AGN will be classified as either type 1 (where we see directly into the nucleus, hence both the broad and narrow line regions are visible) or type 2 (where we see the nucleus through the torus, which hides the BLR but not the narrow line one). Because the torus consists of dust and gas, we expect type 2 AGN to be absorbed in X-rays and optical wavelengths and type 1 not to be. The X-ray absorption is directly measured by the X-ray column density, while the optical one is estimated by means of the colour excess.

We note however that, sometimes, heavily absorbed objects also known as Compton thick AGN, appear “unabsorbed” in X-rays owing to the low statistical quality of the X-ray data and/or the lack of high energy information. To recognize these objects, we can use the diagnostic diagram of Malizia et al. (2007), which plots the X-ray absorption as a function of the source flux ratio F_{(2−10) keV}/F_{(20−100) keV}. For our sample, this is done in Fig. 4, where N_H is for most objects the measured intrinsic absorption (see Table 3) and, for sources PBC J1453.0+2553, PBC J2148.2−3455, and PBC J2333.9−2343, the Galactic one, which is taken here as an upper limit to the X-ray absorption. A clear trend of decreasing flux ratios as the absorption increases is expected and is caused by the 2−10 keV flux being progressively depressed as the absorption becomes stronger. The two lines shown in the figure indeed describe how the flux ratio is expected to change as a function of N_H in the case of objects characterized by an absorbed power law having a photon index of 1.5 and 1.9, respectively. It is evident that most of our sources follow the expected trend with the most absorbed AGN showing progressively smaller F_{(2−10) keV}/F_{(20−100) keV} values. Assuming as a dividing line between absorbed and unabsorbed AGN a column density of 10²² cm^-2 (which is sufficient to hide the BLR of an active nucleus), we note that most of our AGN are above this line or very close to it, as expected given the high percentage of type 2 AGN in our sample. Indeed PBC J0503+2300 and PBC J1453.0+2553, a type 1.5 and a type 1 Seyfert respectively, are both well below the line, while 2 out of 3 Seyfert 1.9 galaxies in our sample have a column density above 10²² cm^-2, i.e. compatible with the idea that these objects are observed through the edge of the torus; the third one, PBC J1546.5+6931, has a column density just below the dividing line between absorbed and unabsorbed objects.

PBC J0814.4+0421, the only NLS1 in the sample, is absorbed, although this is not unusual, as the presence of strong, partial, and/or stratified absorption is one of the two competing models used to explain the complex X-ray spectra of this class of AGN (Panessa et al. 2011).

The only type 1 AGN for which absorption is totally unexpected is PBC J0543.6-2738, which is classified as a type 1.2 AGN, but displays a column density in the range (1.4−5.5) × 10²² cm^-2; it is possible that in this source the gas responsible for the X-ray absorption is highly ionized, instead of being neutral, in which case the accompanying dust would sublimate, yielding a much smaller dust-to-gas ratio and resulting in a reduced optical extinction and consequently in an early type 1 classification. From this perspective, the obscuring material would not be related to the toroidal structure as assumed in the unified theory, but rather to other types of absorption, possibly gas in an outflow from the central nucleus (Winter et al. 2011). Unfortunately, our X-ray spectrum does not have a S/N high enough to enable us to distinguish between ionized and neutral gas models; clearly PBC J0543.6-2738 merits a far more in-depth X-ray study with satellites such as XMM-Newton or Suzaku.

On the other hand, we find two objects that are classified as type 2 Seyferts in optical, but show no absorption in X-rays: PBC J2148.2−3455 and PBC J2333.9-2343. As anticipated above, it is possible that these 2 AGN are Compton thick sources not recognized as such owing to the low quality of the X-ray data; but while this may be the case for the first source, it is certainly not true for the second one, for the following reasons.

PBC J2148.2−3455 is well-studied in X-rays and previous observations strongly indicate that this is indeed a Compton thick AGN. Using Chandra data, Levenson et al. (2005) showed that the active nucleus is probably buried beneath a column density N_Hx ≥ 10²⁴ cm^-2 as indicated by the prominent Fe Kα emission line, which has an equivalent width larger than 1 keV; in addition the F₂₋₁₀ keV/F[Oiii] ratio, which is often used as an alternative way of pinpointing heavily absorbed Seyfert 2 galaxies, is sufficiently small (0.04) to classify the source as a Compton thick object (Bassani et al. 1999). On the other hand, PBC J2333.9-2343, is quite atypical: it lies in the region of type 1 AGN (see Fig. 4, top panel), has a good quality spectrum that provides no indication of an iron line and has a F₂₋₁₀ keV/F[Oiii] ratio of ~1, again similar to type 1 Seyferts (Bassani et al. 1999). The source is peculiar in many other ways: it is listed in the Roma BZCAT as a blazar of unknown type (Massaro et al. 2009), is a flat spectrum radio source (Healey et al. 2007), shows a jet feature in a VLBI 8.4 GHz image, is variable in radio (Ojha et al. 2004), and in X-rays according to a quick look analysis of all publicly available Swift/XRT observations and is finally polarized at radio frequencies (Ricci et al. 2004). This source resembles a nearby blazar, but it has the optical spectral appearance of a type 2 Seyfert, which is quite unusual since flat spectrum radio quasars (one of two types of blazars) are generally broad-line AGN. This is an object that certainly deserves further investigation in X-rays, but also in other wavebands to confirm the above peculiarities.

Fig. 4 Top panel: F2−10 keV/F20−100 keV flux ratio of our sample. Lines correspond to expected values for an absorbed power-law with photon index 1.5 (dotted) and 1.9 (dashed); see text for details. The blue empty symbols indicate sources with Galactic absorption only, while red-filled ones correspond to sources with an extra intrinsic absorption. Bottom panel: X-ray column density versus the optical one computed from the E(B − V) assuming the Galactic extinction law of Cardelli et al. (1989). Triangles are Seyfert 2, squares are Seyfert 1 objects. Points marked with red arrows are upper limits.

Summarizing, we find that the optical versus X-ray classifications for most of our sources broadly fit with the AGN unified scheme, except for a few peculiar objects: PBC J0543.6-2738, which is a type 1 AGN showing some absorption possibly due to outflowing gas, and PBC J2333.9-2343, which is instead a type 2 AGN displaying no absorption and with properties similar to blazars.

In Fig. 4 bottom panel, we plot the X-ray column density versus the optical one (with relative uncertainties), which was measured from the intrinsic colour index E(B − V) using the formula N_Hopt = 3.1 E(B − V) × 1.79 × 10²¹ cm^-2 (Predehl & Schmitt 1995; Rieke & Lebofski 1985). For 5 sources, we decided to take 90% upper limits (marked with red arrows, in Fig. 4) because the errors were larger than the N_Hopt values. Within the unified theory, the X-ray absorption associated with the gas in the torus, which is confined to a region smaller than the narrow line region (NLR); the optical extinction, on the other hand, may come from dust either associated with the torus (internal reddening) or to larger-scale structures such as lanes, bars, or something else (external reddening). As clear from the figure, the majority of our sources has an X-ray column density higher that the optical one. Since most of our objects are AGN of type 2, the optical reddening is related to the NLR, hence is most likely associated with internal rather than external reddening, i.e. the torus; in this case, the bottom panel in Fig. 4 is simply telling us that, with this structure, gas absorption is greater than dust obscuration. This effect was already noticed by Maiolino et al. (2001), who suggested that in AGN the dust-to-gas ratio is much lower than the galactic one or that in the inner parts of the obscuring torus the dust is sublimated by the strong UV radiation field. Another interesting possibility put forward by the same authors is that the dust extinction curve is much flatter than the standard galactic one, for example as a result of the growth of larger dust grains. Only 3 of our objects are located above the 1−1 line implying that there is an optical/(dust) extinction similar to or slightly higher than the X-ray/(gas) absorption. Two of these objects are type 1 AGN and so this is to be expected, as our line of sight to their central nucleus does not intercept the torus; the third object is PBC J1345.4+4141, which is the only type 1.9 AGN of our sample that has an X-ray column density close to but lower than 10²² cm^-2. In this case, it is possible that the optical reddening and the X-ray absorption are unrelated to the torus, but may come from other larger-scale structures. We note that NGC 5290, the optical counterpart of PBC J1345.4+4141, forms a pair of interacting galaxies with NGC 5289 (van Driel et al. 2001); the latter also shows a bright bulge, which is partially hidden by a dark lane and asymmetric absorption (see NED notes). In other words, there is plenty of reddening on large scales to explain the observed properties. It is also possible that in this source the BLR optical continuum has temporarily diminished, leaving visible only a very weak broad Hβ line and leading to the classification as a type 1.9, even if the nucleus is totally unobscured by dust (and also by gas given the not-so-high column density). This is another object for which further observations are clearly encouraged especially at optical/infrared wavelengths.

6. Conclusions

We have either provided for the first time, confirmed, or corrected the optical spectroscopic identifications of 29 sources belonging to the Palermo 39 month Swift /BAT catalogue (Cusumano et al. 2010a). This has been achieved by performing a multisite observational campaign in Europe, South Africa, and Central America.

We have found that our sample is composed of 28 AGN (7 of type 1 and 21 of type 2), with redshifts between 0.008 and 0.075, and 1 CV. Among the extragalactic sources, we found some peculiar objects, such as 3 AGN showing LINER features and 1 object with the properties of a NLS1. For 4 type 1 AGN, we have estimated the BLR size, velocity, and the central black hole mass. We have performed an X-ray spectral analysis of the entire sample and found that overall our sources display X-ray spectra typical of their optical class. More specifically, we have compared the optical to X-ray classifications of our galaxies, to test the AGN unified theory. We found a generally good match between optical class and X-ray absorption, thus validating the unified scheme. However, in a few sources there is a clear discrepancy between the optical and X-ray classifications: PBC J0543.6-2738 is a Seyfert 1.2 displaying mild X-ray absorption, possibly owing to outflowing gas; PBC J1345.4+4141 is instead a Seyfert 1.9 showing no absorption, although its optical class may be related to reddening occurring in large-scale structures or due to a low optical ionization state. More convincingly, outside the unified scheme is PBC J2333.9-2343, which is a Seyfert 2 without an intrinsic X-ray column density; this source has many features that make it very similar to broad-line blazars and yet has only narrow lines in its optical spectrum. Another Seyfert 2 displaying no absorption is PBC J2148.2−3455, but through the use of our diagnostic diagram and information gathered in the literature we conclude that this source is either a Compton thick or heavily absorbed AGN, which is therefore compatible with its optical class. We also compared the X-ray gas absorption with the optical dust reddening for the AGN sample: we find that for most of our sources, specifically those of type 1.9−2, the former is higher than the latter, confirming the early results of Maiolino et al. (2001); this is possibly due to the properties of dust in the circumnuclear obscuring torus of the AGN.

As a final remark, we stress the importance of combining optical with X-ray spectroscopy for hard X-ray selected objects: using information in both wavebands enables us to increase the number of source identifications and classifications, but also perform statistically significant population studies, to understand the physical processes occurring in these objects and study the AGN unified model.

Acknowledgments

We thank Dr. Domitilla de Martino for useful discussions and the referee for comments that helped us to improve the quality of this paper. We also thank Silvia Galleti for Service Mode observations at the Loiano telescope, and both Antonio De Blasi and Ivan Bruni for night assistance at the Loiano telescope. We also thanks Claudia Reyes for night assistance at the ESO NTT telescope. This work is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA. We also acknowledge the use of public data from the Swift data archive. This research has made use of the ASI Science Data Center Multimission Archive, of the NASA Astrophysics Data System Abstract Service, the NASA/IPAC Extragalactic Database (NED), of the NASA/IPAC Infrared Science Archive, which are operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration and of data obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC), provided by NASA’s GSFC. This publication made use of data products from the Two Micron All Sky Survey (2MASS), which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This research has also made use of data extracted from the 6dF Galaxy Survey and the Sloan Digitized Sky Survey archives; the SIMBAD database operated at CDS, Strasbourg, France, and of the HyperLeda catalogue operated at the Observatoire de Lyon, France. The authors acknowledge the ASI and INAF financial support via grants Nos. I/033/10/0, I/009/10/0; P.P. is supported by the INTEGRAL ASI-INAF grant No. 033/1070. L.M. is supported by the University of Padua through grant No. CPDR061795/06. G.G. is supported by FONDECYT 1085267. V.C. is supported by the CONACyT research grants 54480 and 15149 (México). D.M. is supported by the Basal CATA PFB 06/09, and FONDAP Center for Astrophysics grant No. 15010003.

References

All Tables

All Figures

	Fig. 1 Spectrum (uncorrected for the intervening Galactic absorption) of the optical counterpart of the Galactic CV PBC J0826.3-7033.
In the text

	Fig. 2 Spectra (not corrected for the intervening galactic absorption) of the optical counterpart of PBC J0041.6+2534, PBC J0100.6−4752, PBC J0122.3+5004, PBC J0140.4−5320, PBC J0248.9+2627, PBC J0353.5+3713, PBC J0356.9−4040, PBC J0503.0+2300, PBC J0543.6−2738, PBC J0544.3+5905, PBC J0623.8−3212 and PBC J0641.3+3251, PBC J0759.9+2324, PBC J0814.4+0421 and PBC J0919.9+3712.
In the text

	Fig. 3 Spectra (not corrected for the intervening galactic absorption) of the optical counterpart of PBC J0954.8+3724, PBC J1246.5+5432, PBC J1335.8+0301, PBC J1344.2+1934, PBC J1345.4+4141, PBC J1439.0+1413, PBC J1453.0+2553, PBC J1506.6+0349, PBC J1546.5+6931, PBC J1620.3+8101, PBC J2148.2−3455, PBC J2333.9−2343 and PBC J2341.9+3036.
In the text

Fig. 4 Top panel: F2−10 keV/F20−100 keV flux ratio of our sample. Lines correspond to expected values for an absorbed power-law with photon index 1.5 (dotted) and 1.9 (dashed); see text for details. The blue empty symbols indicate sources with Galactic absorption only, while red-filled ones correspond to sources with an extra intrinsic absorption. Bottom panel: X-ray column density versus the optical one computed from the E(B − V) assuming the Galactic extinction law of Cardelli et al. (1989). Triangles are Seyfert 2, squares are Seyfert 1 objects. Points marked with red arrows are upper limits.

In the text